NSD-synthetic
收藏资源简介:
NSD-synthetic数据集是由柏林自由大学等机构的研究人员创建的,包含8名受试者在观看284个精心控制的合成图像时的7T fMRI响应。这些图像被分为8类,包括各种类型的噪声、自然场景、操作过的自然场景、对比度调制、相位一致性调制、单字位置变化、螺旋光栅以及色度噪声变化等。该数据集旨在为视觉处理模型的OOD泛化测试提供支持,以促进更稳健的视觉处理模型的发展,并形成更准确的人类视觉理论。
The NSD-synthetic dataset was created by researchers from Freie Universität Berlin and other institutions. It comprises 7T fMRI responses collected from 8 participants while they viewed 284 meticulously controlled synthetic images. These images are divided into 8 categories, including various types of noise, natural scenes, manipulated natural scenes, contrast modulation, phase consistency modulation, single-word position changes, spiral gratings, and chromatic noise variations. This dataset aims to support out-of-distribution (OOD) generalization tests for visual processing models, advance the development of more robust visual processing models, and foster more accurate theories of human vision.

- 1A 7T fMRI dataset of synthetic images for out-of-distribution modeling of vision柏林自由大学教育心理学系、柏林神经科学中心、柏林计算神经科学伯恩斯坦中心、柏林心智与大脑学校、明尼苏达大学磁共振研究中心、明尼苏达大学神经科学系 · 2025年



